Seizure Prediction Informed by Changes in Circadian Autonomic Modulation

An inexpensive, convenient, and non-invasive seizure detection, monitoring and prediction method

Background

Epilepsy affects 50 million people worldwide and it has highly debilitating medical and social consequences. Although the mechanism behind the seizures is believed to be due to a heightened neuronal activity in the brain cortex, the underlying reasons remain unknown. Nonetheless, there has been great interest and much effort in predicting seizures before they occur. Most prior works have relied on analyzing EEG recordings. However, despite years of research, currently there is no reliable and accurate seizure prediction method. Such a method can enable new types of interventional treatment, reduce the socioeconomic costs of the disease, and dramatically improve the patient’s quality of life.

Technology Overview

This invention includes novel biomarkers that can enable accurate and reliable seizure prediction. The intuition behind the new method is that epilepsy patients exhibit altered sympathetic activity that can be quantified by measuring electrodermal activity and that may be a good biomarker for seizure prediction. The new seizure prediction method relies on long-term recording of electrodermal activity, peripheral body temperature, heart rate, respiratory rate, and blood volume pulse signals. All these signals can be measured using existing wearable sensors. Twenty-four-hour recordings of these signals are analyzed to estimate baseline circadian patterns using nonlinear mixed-effects harmonic models. In addition to the signal level in the time domain, signal power in the frequency domain is computed using a wavelet basis. Deviations from the baseline can be used as indicators of seizure onset and as biomarkers for seizure prediction.

The inventors assessed the accuracy of the new method in and found that the amplitude and spectral power of electrodermal activity was lower in patients with confirmed seizures compared with patients with no seizures. More importantly, they found that the level and spectral power of electrodermal activity was lower than the baseline before and after the seizure, suggesting that electrodermal activity signal can serve as a biomarker for seizure prediction. Electrodermal activity patterns also exhibited significant differences between focal impaired awareness seizures and generalized tonic-clonic seizures.

Further Detail

Vieluf, S., Amengual‐Gual, M., Zhang, B., El Atrache, R., Ufongene, C., Jackson, M.C., Branch, S., Reinsberger, C. and Loddenkemper, T., 2021. Twenty‐four‐hour patterns in electrodermal activity recordings of patients with and without epileptic seizures. Epilepsia, 62(4), pp.960-972.

Benefits

• Inexpensive and convenient
• Follows a multimodal approach by jointly analyzing multiple signals.
• Combines seizure prediction and detection.
• Non-invasive.

Applications

• Seizure prediction.
• Seizure detection and monitoring.
• Discriminating between different types of seizure.
• Assessment of the risk of sudden unexpected death in epilepsy (SUDEP).
• Patient monitoring for other diseases that cause state changes in the autonomic nervous system. 
• General stress management.

IP Status

• Patent application submitted